1

WOMAN KILLING: INTIMATE FEMICIDE IN ONTARIO, 1974 - 1994

Rosemary Gartner University of Toronto

Myrna Dawson University of Toronto

Maria Crawford Women We Honour Action Committee

Major funding for the studies described in this paper was provided by the Ontario Women's Directorate. The Ministry of Community and Social Services and the School of Graduate

2 Studies at the University of Toronto each provided additional funding for one of the studies. The analyses and opinions in the paper are those of the authors and do not necessarily represent the views of any of these funders.

3

WOMAN KILLING: INTIMATE FEMICIDE IN ONTARIO, 1974-1994

Abstract

This paper reviews major findings from our research on women killed by their intimate partners in Ontario. Between 1974 and 1994, killings by intimate partners accounted for between 63% and 76% of all women killed in Ontario. We document trends in intimate femicide, characteristics of victims and offenders, circumstances of the killings, and criminal justice responses to offenders. We also discuss the gender-specific nature of intimate femicides and identify ways in which intimate partner killings by males and females are distinctly different.

4

WOMAN KILLING: INTIMATE FEMICIDE IN ONTARIO, 1974-1994 In March 1988, a young mother of two was killed by her estranged husband in a northern Ontario town. The killer had been visiting his wife who was staying in a shelter for abused women. Convinced that she was not going return to him, he shot her twice at close range. Later that year, in a small town outside of Edmonton, a woman was shot dead in her home by her estranged husband who then shot and killed himself. Miraculously, the woman's three-year old girl, whom she was holding in her arms when she was shot, was not wounded. These women were two of the 202 female victims of homicide in Canada in 1988. They shared with 68 other female victims a marital relationship with their killers. These two women also shared the experience of having been clients and friends of women who worked in shelters for abused women in Ontario. In response to these and other killings of women they had worked with, eight women met in January 1989 to share their experiences and provide each other emotional support. Within a few months the group had named itself the Women We Honour Action Committee, setting itself the task of learning more about the phenomenon of women killed by their intimate partners. With the support of a grant from the Ontario Women's Directorate, they conducted a literature review on women killed by their intimate partners, or intimate femicide. That literature review 1 lead to a number of conclusions about the then-existing state of knowledge about intimate femicide. First, obtaining an accurate estimate of the number of such killings in Canada or in Ontario from statistics in official publications was not possible because official publications restricted their classifications to "spouse killings," which excluded killings

by estranged common-law partners and current or former boyfriends. Second, information on the nature of intimate femicide -- its dynamics as well as its structural and cultural sources -- was incomplete. In part this reflected researchers' reliance on small, highly-select samples, on offenders' recollections of their crimes, and on traditional psychological and psychiatric concepts and classifications. Third, much of the research had been conducted in the United States which is atypical in both the quantity and quality of its homicides. That is, spousal homicides make up a much smaller proportion of total homicides in the U.S. compared to many other nations. Moreover, the ratio of female to male victims of spouse killings is more balanced in the U.S. than in other countries (about 1.3:1, compared to about 3:1 in Canada, Australia, Denmark, the U.K. and other countries). 2 To address these limitations, the Women We Honour Action Committee obtained funding from the Ontario Women's Directorate to conduct a study of intimate femicide in Ontario. The study had three goals: to document for Ontario the incidence of killings of women by intimate partners, including legal spouses, common-law partners and boyfriends, both current and estranged; to describe the characteristics of the people involved in and the circumstances surrounding these killings; and to present the stories of a small number of women who had been killed by their intimate partners. That study, completed in 1992, compiled and analyzed data on all intimate femicides known to authorities in Ontario from 1974 through 1990. 3 A second study, designed to update the data through 1994, was completed in April 1997. 4 In this paper, we describe the major findings of these two studies of intimate femicide. Our purpose is two-fold: first, to provide an overview and statistical picture of intimate femicide

in Ontario for the 21 years from 1974 through 1994; and, second, to locate this statistical picture in what is now a substantially larger and more sophisticated literature on violence against women by intimate partners. That literature encompasses studies similar in many ways to ours -- that is, studies of the incidence and characteristics of relatively large numbers of femicides -- as well as work designed to provide a theoretical and conceptual framework for understanding intimate femicide. We draw on that literature below in discussing our findings.

Framing the Issue of Intimate Femicide After completing our literature review in 1989, we concluded that intimate femicide is a phenomenon distinct in important ways both from the killing of men by their intimate partners and from non-lethal violence against women; and, hence, that it requires analysis in its own right. This view was in contrast to much of the existing literature which treated "spousal violence" as a relatively undifferentiated phenomenon arising out of the intense emotions, stresses, and conflicts that often characteristise marital relations. 5 These analyses tended to locate the sources of "spousal violence" in patterns of learning early in life, in the disinhibitory effects of alcohol consumption, and in dysfunctional patterns of communication between marital partners. Much of this early work also tended to devote limited attention and analysis to gender differences in spousal violence. In response to this neglect of gender, a number of analysts have made gender a central feature of their accounts of spousal violence. Sex role theorists highlight gender differences in socialization which teach males to view toughness, power and control as masculine attributes.

Evolutionary theorists argue that violence is an adaptive strategy for males facing the loss of status and control over their partners. Resource theorists view violence as the ultimate resource available to men when other means of exerting control over their partners are exhausted. General systems theorists argue that for men the rewards of violence against their wives are greater than the costs, because of society's failure to adequately sanction such violence. The arguments of these more gender-sensitive analyses resonated with the experiences of members of the Women We Honour Action Committee. Power, control, and domination were themes that they encountered daily in talking with abused women and that they detected in relationships ending in intimate femicide. In recent work specifically focused on women killed by their intimate partners, these themes have been elaborated and, in the case of feminist analyses, placed in a historical and institutional context. 6 For example, Wilson and Daly cite "male sexual proprietariness" as the predominant motive in the killing of wives across cultures and historical epochs. "Men exhibit a tendency to think of women as sexual and reproductive 'property' that they can own and exchange...Proprietary entitlements in people have been conceived and institutionalized as identical to proprietary entitlements in land, chattels, and other economic resources." They go on to note "[t]hat men take a proprietary view of female sexuality and reproductive capacity is manifested in various cultural practices," including claustration practices, asymmetrical adultery laws, and bride-prices. 7 From this perspective, an extreme, if apparently incongruous manifestation of male proprietariness is intimate femicide. If unable to control or coerce his partner through other means, a man may exert the ultimate control over her by killing her.

Thus, male proprietariness, or male sexual jealousy, has been placed at the centre of many empirical and theoretical analyses of intimate femicide. For example, research on intimate femicide and spousal homicide in Canada, Australia, Great Britain, and the United States 8 has identified a common core in these killings of "masculine control, where women become viewed as the possessions of men, and the violence reflects steps taken by males to assert their domination over 'their' women." 9 This empirical work challenges many of the popular notions about the characteristics of such crimes, for example, the belief that they are explosive, unplanned, and unpredictable acts of passion. At the same time, it contests the validity and coherence of the concept "spousal homicide" with its connotations of sexual symmetry in violence by revealing distinct differences between intimate partner killings by men and those by women. As Dobash et al. note: Men often kill wives after lengthy periods of prolonged physical violence accompanied by other forms of abuse and coercion; the roles in such cases are seldom if ever reversed. Men perpetrate familicidal massacres, killing spouses and children together; women do not. Men commonly hunt down and kill wives who have left them; women hardly ever behave similarly. Men kill wives as part of planned murder-suicides; analogous acts by women are almost unheard of. Men kill in response to revelations of wifely infidelity; women almost never respond similarly... 10

In sum, there have seen significant advances in both empirical and conceptual analyses of lethal violence against women by their partners since the literature review that served as the

impetus for our research. Those advances have not, however, filled all of the gaps identified in our earlier review. In particular, empirical research in Canada has continued to rely largely on official statistics from police sources, which exclude from their classification of spousal homicides killings by men of their estranged common-law partners and girlfriends. Relying on these official statistics also restricts analyses to the information and coding schemes employed by police agencies and personnel. Because of our concerns about the potential for lost information and for the introduction of unknown biases, we relied on a wider range of information sources than typically used in previous research. In this way, our study is unusual in the comprehensiveness of its data. As we see below, it is not however unique in its findings about the nature of intimate femicide.

Data Sources We began our data collection by searching death records kept by the Office of the Chief Coroner for Ontario. Coroner's records provide a centralized source of information on all deaths in Ontario, and a means of identifying and accessing records for deaths identified by the Coroner's Office as homicides. These files frequently contain copies of police reports as well as medical reports on the condition of the body, the way in which the woman was killed and the violence she suffered -- details often not available from other sources. However, coroner's records, like all official sources of information on homicide, are imperfect measures of the actual number of deaths due to homicide. For example, cases of homicide in which no body has been found will not typically appear in coroner's records. As a consequence, we expect our estimates

of the incidence of intimate femicide to undercount the true incidence, an issue we discuss in more depth below. 11 We were able to cross-check and supplement data from coroner's records by reviewing police homicide investigation files for many of our cases. 12 In the second study, we were also able to review data from Crown Attorney files on many of the cases in which charges were laid between 1991 and 1994. In both studies, we supplemented our data from official sources with information from newspaper and magazine articles on some of the killings and on trials of some of the alleged offenders. We compiled this information so that it could be used in both quantitative and qualitative analyses. Our final data collection instrument was designed to provide codes for approximately 52 variables, as well as space to record a narrative of the case where further information was available. 13

The Incidence of Intimate Femicide in Ontario, 1974-1994 Between 1974 and 1994, 1206 women aged 15 and older 14 were killed in Ontario, according to official records. In 1120 (93%) of these cases, the crimes were solved and the killers were identified. In 705 (63%) of the solved cases, the killers were the current or former legal spouses, common-law partners, or boyfriends of their victims. Thus, in Ontario over this 21-year period, intimate partners were responsible for the majority of all woman killings and an average of 34 women were victims of intimate femicide each year. These data indicate that the focus in official publications and some academic research on "spousal homicides" of women

provides an incomplete picture of the more general phenomenon of intimate femicide: excluding killings of women by their estranged common-law partners and current and former boyfriends underestimates the total number of intimate femicides by about 25%. The actual number of intimate femicides in Ontario during these years is undoubtedly higher than this. Intimate partners were certainly responsible for some portion of the cases in which no offender was identified or in which we had too little information to determine the precise nature of the relationship between victim and offender. 15 Adjusting for excluded cases, we estimate that intimate femicides may have accounted for as many as 76% of all femicides in Ontario between 1974 and 1994. However, since it is impossible to know the number and characteristics of excluded cases, the analyses that follow focus only on those 705 cases in which the offender was officially identified as the current or former intimate partner of the victim. Trends in Intimate Femicide. Between 1974 and 1994, the rate of intimate femicide (i.e. the number of victims of intimate femicide per 100,000 women in the general population) ranged from a low of .55 in 1978 to a high of 1.26 in 1991, but appears to follow no particular trend over time (see Figure 1). 16 Dividing the 21-year period in half suggests otherwise, however: the average annual rate for the second half of the period (1.02 per 100,000) was slightly higher than the rate for the first half (.92 per 100,000). [Figure 1 about here] On its own, this difference is insignificant statistically and, it might appear, substantively. However, when compared to the statistically significant decreases in other types of lethal violence, the slightly higher rate of intimate femicide in the latter period takes on greater

importance. The annual rate at which women were killed by strangers or unknown assailants declined significantly from an average of .27 during 1974-1983 to .16 during 1984-1994. Moreover, the annual rate at which men were killed by their spouses also declined significantly, from an average rate of .31 during 1974-1983 to .18 during 1984-1994. In other words, during a period when women's risks from strangers and men's risks from spouses decreased, women's risks from their intimate partners increased slightly. Put another way, after 1984 -- a period of substantial expansion in services for abused women -- men's risks of being killed by intimate partners decreased significantly whereas women's risks did not. Without further analysis of these patterns -- which is beyond the scope of this paper -- we can only speculate as to the reasons for this apparently counter-intuitive finding. One possible explanation is that while the expansion of services for abused women may have resulted in the protection of abusive men from defensive violence by their intimate partners, these same services did not necessarily protect women from their male partners' violence. Research shows that women are most likely to kill their intimate partners after prolonged abuse and when they fear continued or more serious violence against themselves or their children. 17 Where services for abused women are available, women in abusive relationships have an alternative to killing their partners. As Browne and Williams note "By offering threatened women protection, escape and aid, [legal and extra-legal] resources can engender an awareness that there are alternatives to remaining at risk" and thus prevent "killings that occur in desperation." 18 Their analysis of U.S. data lends support to this interpretation: states with higher levels of services to abused women had lower rates of spouse killings of males, but not lower rates of spouse killings of females.

Characteristics of the Victims and Their Killers In many respects, the women killed by their intimate partners and the men who killed them 19 are very similar to women and men in the general population of Ontario, as can be seen from the data in Table 1. For example, women killed by their intimate partners were, on average, about 37 years old; 51% were employed; 80% had children; and 76% were born in Canada. These characteristics do not distinguish the victims from other women in Ontario. [Table 1 about here] In some other respects, however, victims of intimate femicide and their killers differed from women and men in the general population. 20 We can think of these differences as risk markers for intimate femicide because they tell us that some types of women and men face disproportionately high risks of intimate victimization or offending. 21 Each of the markers we discuss below has also been associated with increased risks of lethal violence against women in other research. Relationship Status. Research based on data on spouse killings from Great Britain, Australia, the U.S., and Canada shows that two indicators of the status of the relationship -estrangement and common-law status -- are associated with a higher risk of spouse killings of women. 22 We find similar patterns in our intimate femicide data, although the limited availability of data on marital separation and common-law unions within the general population restricts our analysis somewhat. Census Canada collects information on marital separations, but only for registered

marriages. According to census figures, during the years of our study, 3% of women in Ontario were separated from their legal spouses. According to our data, among the victims of intimate femicide, 16% were separated from their legal spouses. Separation, then, appears to be a risk factor for intimate femicide, since women who were separated from their partners were greatly over-represented among victims of intimate femicide. However, exactly how much greater the risks are for separated women cannot be determined from our data. This is because our measure of separation and the census measure of separation are not precisely comparable: the census measure captures largely long-term and relatively well-established separations, whereas our measure is more sensitive and captures short-term as well as long-term separations. Thus our measure will yield a higher estimate of separated couples. Nevertheless, we expect that even correcting for this difference, we would find separation to be associated with higher risks of intimate femicide. Data on the prevalence of common-law unions in the general population have been collected only since 1991, so we can estimate the risks to women living in common-law relationships only for the most recent years of our research. According to census data, 4% of women were living in common-law unions in 1991 in Ontario. According to our data, during 1991 - 1994 21% of the victims of intimate femicide were killed by common-law partners with whom they were living. Based on our calculations, the rate of intimate femicide for women in common-law unions was approximately six times greater than the average rate of intimate femicide in Ontario in the early 1990s. 23 Clearly, then, women in common-law unions were greatly over-represented among victims of intimate femicide during the early 1990s, and perhaps

in earlier years as well. The higher risks associated with common-law status and estrangement have been interpreted in various ways. Compared to couples in registered marriages, common-law partners are more likely to be poor, young, unemployed, and childless -- all factors associated with higher homicide rates. Compared to co-residing couples, estranged couples are more likely to have a history of domestic violence. 24 This violence may be associated both with women's decisions to leave their relationships and with their greater risks of intimate femicide. In other words, "the fact that separated couples constitute a subset of marriages with a history of discord could explain their higher homicide rates." 25 Male sexual proprietariness could also play a role in the higher risks for common-law and estranged relationships. If, as some have speculated, "husbands may be less secure in their proprietary claims over wives in commonlaw unions than in registered unions," 26 they may be more likely to resort to serious violence to enforce those claims or to lethal violence when those claims are challenged. Echoing a similar theme, several studies that have found elevated risks at separation have cited the male's inability to accept termination of the relationship and obsessional desires to maintain control over his sexual partner: "He would destroy his intimate 'possession' rather than let her fall into the hands of a competitor male." 27 Ethnicity. Women in certain ethnic groups have risks of intimate femicide disproportionate to their representation in the population, according to several studies. For example, in the United States African-American women face unusually high risks of intimate femicide. In Canada, such research is more difficult to do because of restrictions on the

collection of crime statistics by race and ethnicity. However, Statistics Canada has collected data on Aboriginal victims of spousal homicides which indicate that Aboriginal women's rates of spousal homicide are between five and ten times higher than the rates for non-Aboriginal women. 28 We had initially hoped to explore ethnic and cultural differences in the risk of intimate femicide in our research. Our community advisory group, 29 which was composed of women from various ethnic backgrounds active in community organizations, encouraged us to do so. However, our research agreement with the Ministry of the Solicitor General prevented us from compiling "statistics based upon social, cultural, regional, linguistic, racial or ethnic group" from the coroners' records. Nevertheless, we were able to document the number of Aboriginal victims of intimate femicide during these years by relying on other sources of data. 30 We estimate that at least 6% of the victims of intimate femicide in Ontario between 1974 and 1994 were Aboriginal women. Census data for these years indicate that just under 1% of all women living in Ontario classified themselves as Aboriginal. Thus, Aboriginal women in Ontario appear to be over-represented among the victims of intimate femicide. Conversely, Aboriginal men are overrepresented as offenders, since all but four of the Aboriginal victims were killed by Aboriginal men. A number of factors might explain the disproportionate risks of intimate femicide faced by Aboriginal women. Aboriginal Canadians, similar to African-Americans, are an economically impoverished and politically disenfranchised ethnic minority. Considerable research has shown that economic, social, and political disadvantages are associated with higher

homicide rates generally, as well as higher rates of serious spousal violence. In addition, Aboriginal Canadian heterosexual couples also have disproportionate rates of other risk markers for intimate partner violence, such as common-law marital status, low income, bouts of male unemployment, exposure to violence in childhood, alcohol abuse, overcrowded housing conditions, and social isolation -- all of which have been cited as reasons for the higher rates of family violence in Aboriginal communities. 31 Some analysts situate these risk factors within a structural approach that views them as consequences of internal colonialism: "the conditions of colonialism [are] directly related to aboriginal acts of political violence as well as rates of suicide, homicide, and family violence among the aboriginal peoples." 32 Employment. Men's unemployment is commonly cited as a risk factor for wife assaults and is also associated with elevated risks of spousal homicide. Women's employment status, on the other hand, does not appear to be consistently associated with their risks of violence from their partners. 33 The association between men's unemployment and violence against their female partners traditionally has been attributed to the stresses produced by unemployment and limited economic resources. But if this were the case, one would expect to find more evidence that women's unemployment is also associated with spousal violence, which is not the case. For those who see male violence against their partners as one resource for demonstrating power and control, the gender-specificity of the effects of unemployment is not surprising: men who lack more traditional resources (such as economic success) may "forge a particular type of masculinity that centers on ultimate control of the domestic setting through the use of violence." 34

Our data on intimate femicide are consistent with this interpretation. For women, employment status is not associated with differential risks of intimate femicide: 51% of women in both the victim population and the general population were employed during the period of our study. For men, however, employment status is associated with differential risks. Among intimate femicide offenders, 64% were employed, whereas among males in the general population, 73% were employed. In Ontario, then, male unemployment appears to be associated with higher risks of intimate femicide offending. Offenders' Violent Histories. Several studies have shown that men who kill their spouses frequently have histories of violent behaviour, both in and outside of their marital relationships. 35 As Johnson notes, "[a]lthough some wife killings are the result of sudden, unforeseeable attacks by depressed or mentally unstable husbands and are unrelated to a history of violence in the family, most do not seem to fit this description." 36 Because of this, risk assessment tools designed to assess battered women's risk of lethal violence typically include measures of their partners' violence against their children and outside of the home, and threats of serious violence against their wives or others. 37 We also found evidence of unusual levels of violence in the backgrounds of the offenders in our sample. At least 31% of them had an arrest record for a violent offense. 38 At least 53% of them were known to have been violent in the past toward the women they ultimately killed. This corresponds to data for Canada as a whole which indicates that in 52% of spousal homicides of women between 1991 and 1993, police were aware of previous violent incidents between the spouses. 39 In addition, in at least 34% of the cases of intimate femicide, the offenders were

known to have previously threatened their victims with violence. 40 At least 10% of the killings occurred while the offender was on probation or parole, or under a restraining order. It is important to emphasize that these are minimum estimates of the number of offenders with violent and criminal histories. In over 200 of the 705 cases of intimate femicide we did not have enough information to determine if previous violence or police contact had occurred. Nevertheless, the information we were able to find clearly challenges the view that intimate femicides are typically momentary rages or heat-of-passion killings by otherwise non-violent men driven to act out of character by extreme circumstances. A Summary of Risk Markers for Intimate Femicide. Women killed by their intimate male partners and the men who kill them are drawn from all classes, all age groups, all cultural and ethnic backgrounds. However, the victims of intimate femicide and their killers in our study did differ from other women and men in Ontario in some important respects: they were more likely than women and men in the general population to be separated from their partners, to be in common-law relationships, and to be Aboriginal. In addition, men who killed their intimate partners were also more likely to be unemployed and to have histories of criminal violence. These risk markers for intimate femicide have been noted in other research on spousal homicides, and have been interpreted from within various theoretical frameworks. We suggest that they are perhaps most consistent with a framework which views intimate femicide as the manifestation of extreme (if ultimately self-defeating) controlling and proprietary attitudes and behaviours by men toward their female partners.

Characteristics of the Killings An adequate understanding of the sources of intimate femicide will need to take account of the particular characteristics of these killings. Prior research has devoted much less attention to these characteristics than to the characteristics of the individuals involved in the killings. 41 As a consequence, we are limited in both the comparisons we can draw between our findings and the findings from other research and in the interpretations we can offer of these findings. Intimate femicides are typically very private acts: three quarters of the victims were killed in their own homes and, in almost half of these cases, in their own bedrooms. Less than 20% occurred in public places, such as streets, parks, workplaces or public buildings. The most typical method was shooting: one-third of the victims were killed with firearms. Virtually all the other methods required direct and often prolonged physical contact between offenders and their victims: about two-thirds of the offenders stabbed, bludgeoned, beat, strangled, or slashed the throats of their victims. One of the distinguishing features of intimate femicide is the extent and nature of the violence done to the victim. Unlike killings by women of their intimate partners 42 , intimate femicides often involve multiple methods or far more violence than is necessary to kill the victim. For example, in over half of the stabbings, offenders inflicted four or more stab wounds. Beatings and bludgeonings typically involved prolonged violence -- leading some coroners to use the term "over-kill" to describe them. In about 20% of the cases, offenders used multiple methods against their victims, such as stabbing and strangling or beating and slashing. In about 10% of the cases, we also found evidence that the victim's body had been mutilated or

dismembered. The violence in these killings is much more likely to be sexualized than when women kill their intimate partners. 43 Records on approximately half of the cases in our study provided sufficient information for us to determine whether sexual violence was present. In 27% of these cases we found evidence that the victims had been raped, sodomized or sexually mutilated; in another 22% of the cases the victim's body was found partially or completely unclothed. Consumption of alcohol by offenders and/or victims was no more common in intimate femicides than in other killings: 39% of the offenders and 32% of the victims had been drinking immediately prior to the killing. In only 3% of the cases was there evidence of drug use by offenders or victims immediately prior to the killing. Establishing the motives in these killings is fraught with difficulties, as suggested earlier. We made our own determination of the motive after reviewing all the information available to us. In about one-fourth of the cases we felt we had insufficient information to make a judgement about the offender's motive. In the remaining cases, one motive clearly predominated: the offender's rage or despair over the actual or impending estrangement from his partner. This motive characterized 45% of the killings in which we identified a motive. In contrast, women who kill their intimate partners only rarely kill out of anger over an estrangement. 44 Suspected or actual infidelity of the victim was the motive in another 15% of the intimate femicides. In 10% of the cases the killing appears to have been the final act of violence in a relationship characterized by serial abuse. 45 In only 5% of the cases did stressful life circumstances -- such as bankruptcy, job loss, or serious illness -- appear to motivate the

killer 46 ; and in only 3% of the cases was there evidence that the killer was mentally ill. Another feature that distinguishes intimate femicide from intimate partner killings by women is the number of people who die as a result of these crimes. The 705 cases of intimate femicide resulted in the deaths of 977 persons. Most of these additional deaths were suicides by the offenders: 31% of the offenders killed themselves after killing their female partners. 47 But offenders killed an additional 74 persons, most of whom were children of the victims. In addition, over 100 children witnessed their mothers' deaths; thus, while they may have escaped physical harm, they obviously suffered inestimable psychological harm. Our documentation of these characteristics of intimate femicide cannot sufficiently convey the complexity and context surrounding these crimes. Nevertheless, it serves important purposes. Comparing characteristics of intimate partner killings by males and females shows the distinctiveness of these two types of killings -- a distinctiveness that is obscured in studies that treat intimate partner killings by men and women as instances of a single phenomenon. Compared to killings of men by intimate female partners, intimate femicides are much more likely to involve extreme and sexualized violence, to be motivated by anger over separation, to be followed by the suicide of the offender, and to be accompanied by the killing of additional victims. These features highlight the gender-specificity of intimate partner killings and are consistent with a perspective on intimate femicide which views it as based in a larger system of gender[ed] inequality and stratification which perpetuates male control over women's sexuality, labour, and, at times, lives and deaths.

The Criminal Justice Response to Intimate Femicide In our initial study of intimate femicide, we had not intended to collect and analyze data on the criminal justice responses to men who killed their intimate partners -- in part because our primary interest was in the victims of intimate femicide and in part because we did not expect information on criminal justice responses to be consistently reported in coroners' and police records. However, contrary to our expectations, we were able to obtain information on charges laid, convictions, and sentencing in a substantial number of the cases. In 90% of the 490 cases in which we were able to establish that offenders did not commit suicide, we found at least some information on criminal justice processing. In 94% of these cases, the offenders were charged with either first- or second-degree murder 48 . The proportion charged with first- degree murder increased over time, from 34% of the cases in the first half of the period to 52% in the second half. Of the 346 cases for which we found information on dispositions, 10% were disposed through first-degree murder convictions, 35% through second-degree murder convictions, and 38% through manslaughter convictions. Murder convictions increased over time: from 32% of the dispositions in the first half of the period to 56% in the second half. Acquittals accounted for a total of 13% of the cases: 11% were verdicts of not guilty by reason of insanity and 2% were straight acquittals.

Sentencing

information, available for 302 of the men convicted of killing their partners, also indicates that criminal justice responses to intimate femicide increased in severity over time. Prior to 1984, 7% of convicted offenders received no jail time, 14% were sent to secure mental institutions for indefinite periods, 25% were sentenced to less than five years in prison, 38%

were sentenced to between five and ten years, and 15% received sentences of more than 10 years. After 1983, 4% of convicted offenders received no jail time, 7% were sent to secure mental institutions, 10% received sentences of less than five years, 37% received sentences of between five and 10 years, and 41% were sentenced to prison for more than 10 years. This evidence clearly shows that criminal justice responses to intimate femicide became increasingly punitive over the 21 years of our study. How much of this trend reflects increasing punitiveness towards all violent criminals and how much reflects growing public awareness and intolerance of violence against women is an issue requiring further research. 49

The Gender-Specific Nature of Intimate Femicide We have alluded to the gender-specific nature of intimate femicide at various points in our analysis. Here, we develop our ideas about this gender-specificity by considering what is known about gender differences in homicide more generally. We base this discussion on a large body of criminological research on homicide, as well as on data on over 7,000 homicides collected by Rosemary Gartner and Bill McCarthy as part of a separate research project. Among those who study homicide, it is well known that women and men are killed in different numbers, by different types of people, and in different circumstances. Women are less likely to be victims of homicide than men in virtually all societies. Canada and Ontario are no different: men outnumbered women as victims of homicide by a ratio of approximately 2:1 in Canada and in Ontario between 1974 and 1994. This may appear to indicate that women have a sort of protective advantage over men --

that, at least in this sphere of social life, women are not disadvantaged relative to men. However, if we consider gender differences in offending, a different picture emerges. Men accounted for 87% of all homicide offenders in Ontario during these years; and males outnumbered females as offenders by a ratio of almost 7:1. When women were involved in homicides, then, they were almost three times more likely to be victims than offenders; when men were involved in homicides they were more likely to be offenders than victims. In other words, women are overrepresented among victims and under-represented among offenders; for men the opposite is true. Women were also much more likely than men to be killed by someone of the opposite sex, as these figures imply. Fully 98% of all women killed in Ontario between 1974 and 1994 were killed by men. Only 17% of adult male victims were killed by women. Thus, man killing appears to be primarily a reflection of relations within a gender, whereas woman killing appears to be primarily a matter of relations between the genders. Because women are the majority of victims in opposite sex killings, such killings can be seen as one of the high costs to women of male dominance and desire for control in heterosexual relationships. It is in intimate relationships between women and men that male dominance and control are most likely to erupt into physical violence. Women accounted for about 75% of all victims of spouse killings in Ontario during the last two decades. 50 So women outnumber men among victims of spouse killings by a ratio of about 3:1. Moreover, spousal homicides accounted for over 50% of all killings of women but less than 10% of all killings of men. If males, unlike females, are not killed primarily by their intimate partners, who are they killed by and under what circumstances? In Ontario, about 60% of male victims are killed by

acquaintances and strangers; another 20% are killed by unknown assailants. Most male-male homicides are the result of arguments or disputes that escalate to killings. In many cases, both victim and offender have been drinking, and who becomes the victim and who the offender is a matter of happenstance. 51 One classic study of homicide 52 concluded that male-male homicides, as an instance of the more generally physically aggressive behaviour of males, converge with notions of masculine identity. When males kill their intimate female partners, their methods of and motives for killing take on a character distinctive from male-male killings -- a character that denotes the gender specificity of intimate femicide. As noted above, a substantial number of intimate femicides involved multiple methods, excessive force, and continued violence even after the woman's death would have been apparent. 53 The violence in intimate femicides also frequently involves some form of sexual assault, a very rare occurrence in killings of men. The motives in intimate femicide also point to its gender-specificity. The predominance of men's rage over separation as a motive in intimate femicides has no obvious counterpart in killings of men -- even killings of men by their intimate female partners. We agree with others who see this motive as a reflection of the sexual proprietariness of males toward their intimate female partners. In sum, our analysis of intimate femicide and our review of other research and data on gender differences in homicide suggest that woman killing in general and intimate femicide in particular are uniquely gendered acts. By this we mean these killings reflect important dimensions of gender stratification, such as power differences in intimate relations and the

construction of women as sexual objects generally, and as sexual property in particular contexts. Intimate femicide -- indeed, probably most femicide -- is not simply violence against a person who happens to be female. It is violence that occurs and takes particular forms because its target is a woman, a woman who has been intimately involved with her killer.

Conclusion Our purpose in this paper has been to document the incidence and provide a description of the phenomenon of intimate femicide. For some, our approach may be unsatisfying, because we have not proposed a systematic explanation of nor outlined a detailed strategy for preventing these killings. Obviously explaining and preventing intimate femicides are critical tasks, but both require comprehensive knowledge of the phenomenon. The statistical data we have gathered and analyzed are intended to contribute to this knowledge. Nevertheless, we recognize that our overview of the extent and character of intimate femicide in Ontario between 1974 and 1994 has raised at least as many questions as it has answered. Why, for example, did women's risks of intimate femicide increase slightly when public concern over and resources available to abused women were also increasing; when other forms of lethal violence were decreasing; and when criminal justice responses to intimate femicide were becoming more punitive? Why did some women -- such as those in common-law relationships and Aboriginal women -- face disproportionately high risks of intimate femicide? Were there other types of women with elevated risks of intimate femicide -- for example, immigrant women or disabled women -- whom we couldn't identify because of the limitations of

our data? Why are intimate partner killings by men and women so distinctively different? All of these questions deserve answers, but the answers will require research that goes beyond the data and analysis we have been able to present in this paper. There are other types of questions raised by our research that are more immediately pressing, questions about how to prevent intimate femicides. Our research has shown that intimate femicides are not the isolated and unpredictable acts of passion they are often believed to be. Most of the killers in our study had acted violently toward their partners or other persons in the past and many had prior contact with the police as a consequence. Many of the victims had sought help from a variety of sources. In a substantial portion of these intimate femicides, then, there were clear signs of danger preceding the killing, signs that were available to people who might have been able to intervene to prevent the crime. We believe this information could be combined with what we know about the risk factors for intimate femicide -- such as estrangement -- to develop interventions that would save women's lives. This is the question that has been at the core of our research and the recommendation that we tabled at the conclusion of both of our studies. We urged the establishment of a joint forces initiative that would include police, coroners, researchers, experts working in the field, as well as survivors of intimate violence, who would be charged with developing a system to respond more effectively to women when they are at greatest risk of intimate femicide. Such a response would need to be swift and focused on ensuring the victim's safety and deterring the offender from further violence or threats. Of course, this kind of intervention must be coupled with efforts to address the

underlying sources of intimate femicide. If, as we and others have argued, the sources lie at least in part in attitudes and behaviours that have been supported for centuries by patriarchal systems of power and privilege, those attitudes and behaviours, as well as the systems supporting them, must be confronted and contested. Some feminists argue that one means of doing this is through refining and reformulating law as a weapon against men's intimate violence against women. Isabel Marcus, for example, argues for identifying domestic violence as terrorism and, as such, a violation of international human rights accords. 54 Elizabeth Schneider suggests redeploying the concept of privacy, not to keep the state out of intimate relationships as the concept has been used in the past, but to emphasize individuals' autonomy and independence. She argues this affirmative aspect of privacy could frame a new feminist agenda against woman abuse. 55 As these and other analyses emphasize, preventing intimate femicides will require that the public as well as those working in fields relevant to the prevention of violence begin to see intimate femicide as a preventable crime. From our own and others' research on intimate violence, it should be apparent that these crimes are patterned and predictable. The danger lies in maintaining the view that violence is inevitable, unavoidable, and inherent in intimate relationships. Such fatalism must be challenged, so that women's safety in and outside their homes is seen as an achievable and preeminent goal.

ENDNOTES 1 Women We Honour Action Committee and Rosemary Gartner, Annotated Bibliography of Works Reviewed for Project on Intimate Femicide (Toronto: Women We Honour Action Committee, 1990). 2 For efforts to analyze the source of these variations in the sex ratio of spouse killings, see Margo Wilson and Martin Daly, "Who kills whom in spouse killings? On the exceptional sex ratio of spousal homicides in the United States," Criminology 1992, 30:189-215; Wendy Regoeczi and Robert Silverman, "Spousal homicide in Canada: Exploring the issue of racial variations in risk," paper presented at the 1997 Annual Meetings of the American Society of Criminology. 3 Maria Crawford, Rosemary Gartner, and the Women We Honour Action Committee, Woman Killing: Intimate Femicide in Ontario, 1974-1990 (Toronto: Women We Honour Action Committee, 1992). 4 Maria Crawford, Rosemary Gartner, and Myrna Dawson in collaboration with the Women We Honour Action Committee, Woman Killing: Intimate Femicide in Ontario, 1991-1994 (Toronto: Women We Honour Action Committee, 1997). 5 See, for example, Peter D. Chimbos, Marital Violence: A Study of Interspousal Homicide (San Francisco: R & E Associates, 1978); Martin Blinder, Lovers, Killers, Husbands, and Wives (New York: St. Martin's Press, 1985); James Boudoris, "Homicide and the family," Journal of Marriage and the Family 1971, 32:667-676; William Goode, "Violence among intimates," Pp. 941-977 in Crimes of Violence [Vol 13], edited by D. Mulvihill and M. Tumin (Washington D.C.: USGPO, 1969). 6 See, e.g., Jacquelyn C. Campbell, "'If I can't have you no one else can': Power and control in homicide of female partners," pp. 99-113 in Femicide: The Politics of Woman Killing, edited by J. Radford and D.E.H. Russell (New York: Twayne, 1992); Govind Kelkar, "Women and structural violence in India," pp. 117-123 in Femicide: The Politics of Woman Killing, edited by J. Radford and D.E.H. Russell (New York: Twayne, 1992); Isabel Marcus, "Reframing 'domestic violence': Terrorism in the home," pp. 11-35 in The Public Nature of Private Violence: The Discovery of Domestic Abuse, edited by M.A. Fineman and R. Mykitiuk (New York: Routledge, 1994); Martha A. Mahoney, "Victimization or oppression? Women's lives, violence, and agency," pp. 59-92 in The Public Nature of Private Violence: The Discovery of Domestic Abuse, edited by M.A. Fineman and R. Mykitiuk (New York: Routledge, 1994). 7 Margo Wilson and Martin Daly, "Til death do us part," pp. 85 in Femicide: The Politics of Woman Killing, edited by J. Radford and D.E.H. Russell (New York: Twayne, 1992). 8 See, e.g., Allison Wallace, Homicide: The Social Reality (New South Wales: NSW Bureau of

Crime Statistics and Research, 1986); Kenneth Polk, When Men Kill: Scenarios of Masculine Violence (Cambridge: Cambridge University Press, 1994); Martin Daly and Margo Wilson, Homicide (New York: Aldine de Gruyter, 1988); P.W. Easteal, Killing the Beloved: Homicide Between Adult Sexual Intimates (Canberra: Australian Institute of Criminology); R. Emerson Dobash and Russell P. Dobash, "The nature and antecedents of violent events," British Journal of Criminology 1984, 24: 269-288. 9 Polk (1994), p. 56. 10 Russell P. Dobash, R. Emerson Dobash, Margo Wilson, and Martin Daly, "The myth of sexual symmetry in marital violence," Social Problems 1992, 39:81. 11 Coroner's records are limited in another obvious and unavoidable way: they are observations removed in time and space from the actual killing. As a consequence, the description in the records will be shaped by the interests and perspectives of the observer. A coroner's perspective is that of an investigator after the fact, and his/her primary interest is in determining the cause and means of death. Thus, the information recorded by coroners is intended to serve these purposes, not the interests of researchers. 12 Different procedures were used in the two studies to obtain access to municipal police and OPP records. These records are not centrally compiled and it was impossible to contact and obtain cooperation from all of the forces around Ontario which investigate and keep records on cases of homicide. 13 Obviously, the coded data provide only a partial and, in some respects, an incomplete portrayal of intimate femicide. The lives and deaths of the women represented in these statistics cannot be sufficiently understood from counts and categorizations. For this reason, we devoted a considerable portion of our first study to reconstructing the stories of some of the women who died through interviews with their family and friends.

14. Our research has looked only at killings of females aged 15 and older because the killing of children differs in distinctive ways from the killing of adults. 15. The number of intimate femicides is under-counted in official records for other reasons as well. For example, in some cases of intimate femicide, the woman's death may be incorrectly classified as due to suicide, accident, or natural causes. Among the intimate femicides in our study, at least eight were not initially classified as homicides and only re-classified after further investigation. Another example of this occurred while this article was being written: the body of a southern Ontario

woman who died by hanging was exhumed and an investigation revealed she had not killed herself, as originally determined, but had been killed by her boyfriend. 16 Although there are no statistics on the rate of intimate femicide for Canada as a whole, there are statistics on the rate of spousal killings of women. Since the mid-1970s, trends in Ontario's rate have paralleled those for Canada as a whole; and the mean rate for Ontario (.77) is very close to the mean rate for Canada (.83). 17 See, e.g., Angela Browne, When Battered Women Kill (New York: The Free Press, 1987). 18 Angela Browne and Kirk Williams, "Exploring the effect of resource availability on the likelihood of female-perpetrated homicides," Law and Society Review 1989, 23:75-94. 19 Of the cases of intimate femicide between 1974 and 1994, we found only three in which the offender was a woman. 20. Identifying differences between victims or offenders and women and men in the general population requires establishing the proportion of victims (or offenders) with the particular characteristic and comparing this to the proportion of women (or men) in the general population of Ontario during the years 19741994 with the same characteristic. If the former proportion is larger than the latter proportion, this indicates that women with that particular characteristic are over-represented among victims of intimate femicide. Tests for statistically significant differences are not appropriate here because the data are based on populations, not samples. Because we used information from census reports to determine the characteristics of women in the general population of Ontario, we were limited in our search for risk markers of intimate femicide to characteristics which are measured in the census. 21 By highlighting these characteristics, we do not mean to obscure the fact that women from all types of backgrounds and in all types of relationships are victims of intimate femicide; nor do we mean to imply that certain characteristics of women make them likely targets for intimate violence. Rather, we would suggest that certain groups of women may be more vulnerable to intimate violence because they share characteristics that have isolated them, limited their access to resources for protection, or prevented them from obtaining a level of personal security that many Canadians take for granted. 22 Holly Johnson, Dangerous Domains: Violence Against Women in Canada (Toronto: Nelson Canada, 1996); Margo Wilson and Martin Daly, "Spousal homicide risk and estrangement," Violence and Victims 1993, 8:3-16; Wallace (1986); Campbell (1992).

23. The average annual rate of intimate femicide (per 100,000 women aged 15 and older) for the years 1991 - 1994 was calculated by: 1) dividing the number of victims during those years (159) by the number of women aged 15 and older in the Ontario population in 1991 (4,130,450); 2) multiplying this figure by 100,000; and 3) dividing this figure by four (the number of years). This yields an average annual rate of .96 per 100,000 women aged 15 and older. The average annual rate of intimate femicides of women living in common-law unions was calculated by: 1) dividing the number of victims living common-law during those years (45) by the number of women aged 15 and older in Ontario living in common-law unions in 1991 (182,155); 2) multiplying this figure by 100,000; and 3) dividing this figure by 4. This yields an annual average rate of 6.18 per 100,000 women aged 15 and older living in common-law unions. 24 Holly Johnson and Vincent Sacco, "Researching violence against women: Statistics Canada's National Survey," Canadian Journal of Criminology 1995, 3:281-304; Karen Rodgers, Wife Assault: The Findings of a National Survey (Ottawa: Canadian Centre for Justice Statistics, 1994). 25 Margo Wilson and Martin Daly, "Spousal homicide," Juristat Service Bulletin 1994, 14:8. 26 Margo Wilson, Holly Johnson and Martin Daly, "Lethal and nonlethal violence against wives," Canadian Journal of Criminology 1995, 37:343. 27 Polk (1994), p. 29; see, also, Christine Rasche, "Stated and attributed motives for lethal violence in intimate relationships," paper presented at the 1989 Annual Meetings of the American Society of Criminology; Wilson and Daly (1993); Campbell (1992). 28. Leslie W. Kennedy, David R. Forde, and Robert A. Silverman, "Understanding homicide trends: Issues in disaggregating for national and cross-national comparisons," Canadian Journal of Sociology 1989, 14:479-486; Robert Silverman and Leslie Kennedy, Deadly Deeds: Murder in Canada (Toronto: Nelson Canada, 1993). 29. This group was formed at the beginning of our first study and met with the principal researchers regularly to review the research for cultural sensitivity and validity. At the completion of the first study, its members also reviewed and made contributions to the final report. 30.

The final report for our first study (pp. 67-76) documents

the problems with collecting information on race and cultural backgrounds of crime victims and offenders, as well as the procedures we followed to gather the data on Aboriginal victims. 31 Health and Welfare Canada, Reaching for Solutions: Report of the Special Advisor to the Minister of National Health & Welfare on Child Sexual Abuse in Canada. (Ottawa: Supply & Services, 1990); David A. Long, "On violence and healing: Aboriginal experiences, 1960-1993," pp. 40-77 in Violence in Canada: Sociopolitical Perspectives, edited by J.I. Ross (Don Mills: Oxford University Press, 1995). 32 Long (1995), p. 42. See also Sharlene Frank, Family Violence in Aboriginal Communities: A First Nations Report (British Columbia: Report to the Government of British Columbia, 1993); Ronet Bachmann, Death and Violence on the Reservation: Homicide, Family Violence, and Suicide in American Indian Populations (New York: Auburn House, 1993) 33 H. Johnson (1996); Ross Macmillan and Rosemary Gartner, "Labour force participation and the risk of spousal violence against women," paper presented at the 1996 Annual Meetings of the American Society of Criminology; Gerald Hotaling and David Sugarman, "An analysis of risk markers in husband to wife violence: The current state of knowledge," Violence and Victims 1986, 1:101-124; Merlin Brinkerhoff and Eugene Lupri, "Interspousal violence," Canadian Journal of Sociology 1988, 13:407-434. 34 James W. Messerschmidt, Masculinities and Crime: Critique and Conceptualization of Theory (Lanham, MD: Rowman and Littlefield, 1993), p. 149. 35 See, e.g., review in Johnson (1996), pp. 183-186. 36 Johnson (1996), p. 183. 37 See, e.g., Jacquelyn Campbell, "Prediction of homicide of and by battered women," pp. 96113 in Assessing Dangerousness: Violence by Sexual Offenders, Batterers, and Child Abusers (Thousand Oaks, CA: Sage, 1995). 38 Another 30% had been arrested and charged with non-violent criminal offenses. 39 Canadian Centre for Justice Statistics, Homicide Survey, unpublished statistics, 1993. 40 In contrast, in only 6% of the cases were the victims known to have been violent toward their killers in the past; and in only 2% of the cases were the victims known to have previously threatened their partners with violence. 41 What researchers can describe about homicide and femicide is largely determined by the types of information officials collect. This means that many details about the events leading up

to the killing, the dynamics of the interaction immediately preceding the killing, or the states of mind of victim and offender are absent or at most only hinted at in official reports. Some characteristics of intimate femicide can be easily and reliably determined, such as where they occurred or whether weapons were involved. Other characteristics -- such as the offender's motivation -- are more susceptible to post hoc reconstructions that introduce the inevitable biases of observers and officials. When we collected and coded information, we reviewed all the information available to us and made our own best judgements about these characteristics. We recognize, however, that our judgements are necessarily based on limited information about extremely complex events. Our discussion of the characteristics of the killings therefore should be viewed with these limitations in mind. 42. We base this and other conclusions about the characteristics of intimate partner killings by women on data from an on-going study by the first author of over 7,000 homicides in two Canadian cities and two U.S. cities over the twentieth century. 43. Indeed, none of the data or research with which we are familiar indicates that women who kill their intimate partners exact sexual violence against their victims. 44.

Browne (1987); Daly and Wilson (1988).

45. This does not mean that offenders who appeared to act for other motives had not engaged in systematic abuse of the women they killed. Rather, it indicates that in 10% of the cases, the only motive we could identify was systematic, serial abuse that ultimately lead to the woman's death. 46 Typically, offenders who kill under these circumstances are characterized as extremely depressed, and are more likely than other offenders to commit or attempt suicide after the killing. Nevertheless, some have argued that sexual proprietariness can still be seen in killings apparently motivated by stressful life circumstances (e.g. Daly and Wilson, 1988). According to this view, when men kill their wives (and often their children, as well) because they feel they can no longer provide for them, their acts suggest that they see their wives as possessions to dispose of as they see fit and/or that they cannot conceive of their wives having an existence separate from their own. 47 Other research has noted the high rates at which offenders suicide after intimate femicides, and has contrasted this to the rarity of suicides by women who kill their intimate partners (see, e.g. Carolyn R. Block and A. Christakos, "Intimate partner homicide in Chicago over 29 years," Crime and Delinquency 1995, 41: 496-526. Daly and Wilson (1988) have suggested that this pattern is grounded in males' feelings of possessiveness and ownership over their partners. 48 Murder is first-degree when the killing is planned and deliberate, when the victim is an

officer of the law, or when a death is caused while committing or attempting to commit another offense, such as kidnapping. Any murder that does not fall within these categories is seconddegree murder. According to the courts, the distinction between first and second degree murder is made solely for sentencing purposes. While anyone convicted of murder is setenced to imprisonment for life, the parole ineligibility period varies between first- and second-degree murder.

49 Some analysts (e.g. Elizabeth Rapaport, "The death penalty and the domestic discount," pp. 224-251 in The Public Nature of Private Violence: The Discovery of Domestic Abuse, edited by M. Fineman and R. Mykitiuk (New York: Routledge, 1994) have speculated that the killing of a woman by her intimate male partner is treated more leniently by the criminal justice system than other types of homicides, such as killings of men by female intimate partners. However, empirical evidence in this area is sparse and not conclusive. 50. We use the category 'spouse killings' here because we could find no statistics on the number of men killed by intimate partners, only statistics on men killed by spouses. To be comparable, we compare these figures to the number of women killed by spouses -- a subset of all intimate femicides. 51. Marvin Wolfgang has noted in Studies in Homicide (New York: Harper and Row, 1967) that where males are victims of homicide, victim-precipitation of the violence is fairly common. 52. Marvin Wolfgang, Patterns in Criminal (Philadelphia: University of Pennsylvania Press, 1958).

Homicide

53. Wolfgang (1967) found a similar pattern in his study of homicides in Philadelphia. 54 Marcus (1994). 55 Elizabeth Schneider, "The violence of privacy," pp. 36-58 in The Public Nature of Private Violence: The Discovery of Domestic Abuse, edited by M. Fineman and R. Mykitiuk (New York: Routledge, 1994).